# Pass 9 — Cognitive debt review

Date: 2026-05-16

Specific, actionable consolidations and archives only. Every recommendation names tasks and a verification — no vague "this could be cleaner".

## CONSOLIDATE (4 clusters)

### CONS-1: Trader-picker cluster

**Tasks:** shared-user-dropdown (P1), multi-trader-scorecards (priority=null → bump to P1 per drift), phase-4-trader-selector (P1), monthly-per-trader-view (P2)
**Proposed action:** keep all 4 but mark them as a single sequenced flow with `depends_on` chains.
**Why not collapse to 1:** each represents a distinct deliverable on a different property (foundation/scorecards/monthly_reports). Collapsing would lose property-attribution.
**Combined done_when (clearer than originals):**
- `shared-user-dropdown` rendered on /scorecards/, /monthly/, /strategies/, /accountability/
- `multi-trader-scorecards` produces non-empty scorecards for 5 traders
- `phase-4-trader-selector` is the same dropdown component
- `monthly-per-trader-view` regenerates monthly per ?user= URL param

Verdict: **clarity gained** via the explicit dep chain. Patch.json wires the chain.

### CONS-2: Strategy codification leaf cluster

**Tasks:** strategy-vpi-rules + strategy-backside-rules (the two leaf nodes)
**Proposed action:** keep separate (different strategies); but bundle the registry edits into ONE PR. Both write to `/system/strategies.json`.
**Why:** registry edits to the same file have inherent conflict risk if shipped separately; bundling them removes a sequencing hazard.
**Combined done_when:** `/system/strategies.json` has codified rules + gates + thresholds for vwap-pull-in-long, vwap-pull-in-short, backside-scalp-long, backside-scalp-short; classifier output matches expected tag distribution on a reference day.

Verdict: **clearer than originals** because it documents the shared file edit.

### CONS-3: risk_hud build cluster (under decision)

**Tasks:** risk-hud-stop-gauge, risk-hud-summary-row, risk-hud-sort-risk, risk-hud-auto-refresh, risk-hud-api-wire
**Proposed action:** keep all 5 but consolidate behind `risk-hud-accountability-hud-dedup-review` (P1, user-owned) as a decision gate. After dedup decision:
- If risk_hud is retained: 5 tasks remain and can ship in order.
- If risk_hud is retired in favor of acct-v2: all 5 cancel together.

Patch.json adds `blocked_by=["risk-hud-accountability-hud-dedup-review"]` to all 5.
Verdict: **operational confusion eliminated** (no more "should I work on risk_hud or not?" ambiguity).

### CONS-4: Webhook enrichment leaf cluster (downstream of `wire-webhooks`)

**Tasks:** remove-na-labels (vague), smb-rubric-grading (no DoD), overtrading-detect (concrete), per-trade-charts (concrete), strategy-followed-plan (joins to mgp-sip-db)
**Proposed action:** keep separate but mark `wire-webhooks` as the single upstream gate (already done in patch.json). Reframe `remove-na-labels` from VAGUE_CANDIDATE → "scorecards no longer show 'webhook-data N/A' anywhere for trades with an indicator_alert row".
**Verdict:** modest clarity gain; the cluster is correctly structured already.

## ARCHIVE (3)

Strict criteria: task is "low-priority + old + zero mentions in last 14 daily writeups + no demand evidence". Drag evidence shown for each.

### ARCH-1: locate-bitcoin-todos (P2, user, created 2026-05-06)

**Drag evidence:**
- Exhaustive search already done 2026-05-06 (filename + content; per task note: "All hits are legitimate non-TODO files: KITE backtest summary...")
- 10 days no progress, no nightly writeup mention
- User-owned, P2

**Recommendation:** archive with closure note: "Exhaustive search performed 5/6 found no Bitcoin TODO doc on Z drive. If James remembers a separate location, create a fresh tracked task; this one is closed for absence of evidence."

### ARCH-2: hot-key-managment-sell-of-postion (P2, claude, created 2026-05-12)

**Drag evidence:**
- Property `alerts-hotkeys-etc` is the inactive property (no completed tasks; only 1 pending = this one)
- Phase = `design`, no design produced in 4 days
- Note is wishful: "I'm thinking this can help with overtrading..."
- Conflicts with overtrading-detect (P2-ish, claude, scorecards) which exists for the same outcome

**Recommendation:** archive with closure note: "Outcome (reduce overtrading via partial exits) belongs in `overtrading-detect` once webhooks flow. Re-open under that task or as a child once Phase 2 webhook enrichment lands."

### ARCH-3: rescue-localstorage (no-priority, user, created 2026-04-18)

**Drag evidence:**
- 28 days no progress
- Outcome (recover localStorage scorecard_journal_* keys before DB cutover) is already achieved by rescue-0416 (completed 4/18 — note: "Rescue 4/16 IBIT notes to durable file")
- Once `journal-user-steps` ships, new journal writes go to DB; the localStorage rescue is a one-time backfill that may already be done implicitly

**Recommendation:** confirm with James, then archive. If James has notes from 4/17-onward that were lost, keep the task; otherwise it's chasing a hypothetical loss.

## SIMPLIFY (3)

### SIMP-1: pm-nightly-task

**Current state:** Note describes 5-part deliverable (Z scan + Claude summary + writeup + auto-update todos + visual progress). 3 of 5 are shipped per the STATUS bump.
**Simplification:** Split into 2 tasks:
- "Nightly PM: auto-update todos.json on detected completions" (the residual)
- "Nightly PM: visual progress section in writeup" (the other residual)

Each can ship independently. Verdict: **clarity gained** for the EXECUTION_READY check.

### SIMP-2: pm-nightly-silent-failure-detection

**Current state:** Concrete diagnosis (5/7 silent failure documented).
**Simplification:** None needed — already well-scoped. The note is the diagnosis.

### SIMP-3: ux-cross-nav

**Current state:** "every property must link to every other property" — affects 8 properties.
**Simplification:** Phase: ship shared `_shared/nav.html` first, then include on properties one at a time. Patch.json adds a `done_when` listing the 8 properties so the work is incrementally verifiable.

## NARROW_FOCUS (2)

### NARR-1: today_focus is doing the job

The `meta.today_focus` block on todos.json carries the 3-task priority list for the trading week:
1. mgp-drop-vital-independent-source (in_progress) — Monday cutover
2. strategies-family-migration (NOT a pending task — happens to match the in_progress's note about ripple checks; appears to be informational)
3. mgp-cron-skipped-5-14 (P0, owner=James)

Confirmed: today_focus[1] (`strategies-family-migration`) does **not** resolve to a pending task id. It's a label for "validate the 5/15 migration" — the actual verification is in `meta.today_focus[1].why`. **Not a bug**, but worth knowing: the today_focus block can carry validation labels that aren't task ids.

### NARR-2: current_priority_id discipline

`current_priority_id` = `acct-v2-bug-risk-bump-ignores-date-range` (P0, jimbo, 9 days old). James reviews this as the single "what comes next" pointer. Healthy mechanism.

## Cognitive debt summary

- **4 consolidations** (none collapses tasks; all add explicit sequencing or shared done_when).
- **3 archives** with drag evidence (1 immediate, 1 confirm-then-archive, 1 awaits user confirm).
- **3 simplifications** (split pm-nightly-task; ship ux-cross-nav phased).
- **0 destructive collapses.** The system's task granularity is appropriate.

Net cognitive-debt-load reduction if all applied: ~4 tasks closed/archived, ~5 tasks gain clarity via dependency chains, 1 ambiguous architecture decision (risk_hud) becomes a single decision rather than 5 build tasks. **Modest but real cleanup.**

The biggest cognitive-debt insight is the **catalog drift** documented in drift.md — that's where the highest-leverage clarity gain lives, and it's covered by the headline weekend pick recommendation.
